| 1 |
On Sep 26 2013 4:21 AM, Brian Dolbec wrote: |
| 2 |
> On Wed, 2013-09-25 at 23:29 -0600, EBo wrote: |
| 3 |
>> On Sep 25 2013 2:58 PM, Jauhien Piatlicki wrote: |
| 4 |
>> ... |
| 5 |
>> It would interesting if this work could work with other efforts to |
| 6 |
>> tackle specific other specific backends like R and the R_Overlay |
| 7 |
>> project. |
| 8 |
> |
| 9 |
> R needed more fine grained control over the ebuilds generated, hence |
| 10 |
> R_Overlay, with it's database of rules/overrides used to generate the |
| 11 |
> ebuilds. In fact for the pypi backend some code inspiration is taken |
| 12 |
> from R-overlay to produce a database which can/will be downloaded to |
| 13 |
> speed up overlay generation/update (that feature may not be complete |
| 14 |
> yet). |
| 15 |
|
| 16 |
Good to know. |
| 17 |
|
| 18 |
>> > ... |
| 19 |
>> > 2. Emerge app-portage/g-sorcery (note, it depends on |
| 20 |
>> > app-portage/layman-9999) |
| 21 |
>> |
| 22 |
>> With Gentoo's distain for live ebuilds, do you have a sense for |
| 23 |
>> which |
| 24 |
>> version of layman might support the necessary functionality? |
| 25 |
> |
| 26 |
> |
| 27 |
> I will include it in the next release of layman. But none of the |
| 28 |
> current releases have it. I have to make some changes for the new |
| 29 |
> portage config changes now in the newer portage versions. So, it |
| 30 |
> shouldn't be too long. |
| 31 |
> |
| 32 |
> And what's so hard about: |
| 33 |
> |
| 34 |
> echo "=app-portage/layman **" >> /etc/portage/package.accept_keywords |
| 35 |
> emerge layman |
| 36 |
> |
| 37 |
> and for easier live ebuild updates: |
| 38 |
> |
| 39 |
> emerge app-portage/smart-live-rebuild |
| 40 |
|
| 41 |
it is not an issue with it being hard, but stability and consistency -- |
| 42 |
the code that I am you checked out of a repository and built last week |
| 43 |
is not necessarily the same code I checked out and emerged today. |
| 44 |
Someone might have made a change the day before yesterday which |
| 45 |
introduced a bug, and when I post a bug report there is no way to know |
| 46 |
what code either of us are compiling. |
| 47 |
|
| 48 |
I always preferred using date-stamped ebuilds which check out a |
| 49 |
particular revision of a code base. So I know that on 20130923 this is |
| 50 |
when I tested revision #219 with a hashtag of "fa1f0b5c4db5". Then when |
| 51 |
I go back to test or depend on something, I know exactly what version we |
| 52 |
are talking about. |
| 53 |
|
| 54 |
>> > ... |
| 55 |
>> recently I started looking at OSv, and it uses gradle. The current |
| 56 |
>> gentoo ebuilds I have found are -bin installs. Since OSv has a Java |
| 57 |
>> VM |
| 58 |
>> as one of its primary pieces, it might be both instructive and |
| 59 |
>> useful to |
| 60 |
>> contact them and get the skinny on why they use gradle since you are |
| 61 |
>> looking at Clojure. |
| 62 |
>> |
| 63 |
>> I mentioned rubygems/ruby above. Maybe that would be an interesting |
| 64 |
>> avenue also. |
| 65 |
> |
| 66 |
> I don't know ruby/rubygems, but I have a feeling it might be better |
| 67 |
> off |
| 68 |
> using R_overlays code to generate a hosted Ruby_overlay |
| 69 |
|
| 70 |
sounds about right. Since you were asking for suggestions I thought I |
| 71 |
would oblige ;-))) |
| 72 |
|
| 73 |
Anyway, thanks again for the great work, and I hope that this and the |
| 74 |
R_Overlay continues to grow and progress. |
| 75 |
|
| 76 |
Best regards, |
| 77 |
|
| 78 |
EBo -- |
Archives